Tilburg University
Diurnal cortisol and decision making under risk in problem gambling.
Buchanan, Tony W.; Mcmullin, Sara D.; Mulhauser, Kyler; Weinstock, Jeremiah; Weller, Joshua A.
Published in:
Psychology of Addictive Behaviors DOI:
10.1037/adb0000474 Publication date: 2020
Document Version Peer reviewed version
Link to publication in Tilburg University Research Portal
Citation for published version (APA):
Buchanan, T. W., Mcmullin, S. D., Mulhauser, K., Weinstock, J., & Weller, J. A. (2020). Diurnal cortisol and decision making under risk in problem gambling. Psychology of Addictive Behaviors, 34(1), 218-229. https://doi.org/10.1037/adb0000474
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal
Take down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
White Rose Research Online URL for this paper: http://eprints.whiterose.ac.uk/166692/
Version: Accepted Version
Article:
Buchanan, TW, McMullin, SD, Mulhauser, K et al. (2 more authors) (2020) Diurnal cortisol and decision making under risk in problem gambling. Psychology of Addictive Behaviors, 34 (1). pp. 218-229. ISSN 0893-164X
https://doi.org/10.1037/adb0000474
©American Psychological Association, 2020. This paper is not the copy of record and may not exactly replicate the authoritative document published in the APA journal. Please do not copy or cite without author's permission. The final article is available, upon publication, at: https://doi.apa.org/doi/10.1037/adb0000474
eprints@whiterose.ac.uk https://eprints.whiterose.ac.uk/
Reuse
Items deposited in White Rose Research Online are protected by copyright, with all rights reserved unless indicated otherwise. They may be downloaded and/or printed for private study, or other acts as permitted by national copyright laws. The publisher or other rights holders may allow further reproduction and re-use of the full text version. This is indicated by the licence information on the White Rose Research Online record for the item.
Takedown
If you consider content in White Rose Research Online to be in breach of UK law, please notify us by
Diurnal Cortisol and Decision Making Under Risk in Problem Gambling
Tony W. Buchanan, Sara McMullin, Kyler Mulhauser, & Jeremiah Weinstock Saint Louis University
Joshua A. Weller Tilburg University
Corresponding Author: Dr. Tony W. Buchanan Department of Psychology Saint Louis University 3700 Lindell Blvd. St. Louis, MO 63108 Tbuchan7@slu.edu
Author Note
Abstract
The aim of this study was to assess the influence of diurnal cortisol profile on decision making under risk in individuals with problem gambling and a healthy control group. We examined the relationship between diurnal cortisol, assessed over the course of two days, and a battery of tasks that assessed decision making under risk, including the Columbia Card Task and the Cups Task. Thirty individuals with problem gambling and 29 healthy individuals took part in the study. Those with problem gambling showed blunted diurnal cortisol and more risk taking behavior compared to those in the healthy control group. Blunted cortisol profile was associated with more risky behavior and less sensitivity to losing money in problem gambling. These findings suggest that blunted stress physiology plays a role in specific parameters of risky decision making in problem gambling.
Diurnal Cortisol and Decision Making Under Risk in Problem Gambling
Gambling disorder is the most common behavioral addiction and frequently co-occurs with other psychiatric disorders, damaging health and financial stability (Petry, Stinson, & Grant, 2005; Potenza, 2015). Abnormal stress physiology serves as a risk factor for the development of other addictions such as substance use disorders (Biback & Zack, 2015; Lovallo, 2006) and is related to greater addiction severity, cravings, and poor treatment outcomes (Buchanan & Lovallo, 2018; Lovallo, 2006). The little work done on stress physiology in gambling disorder suggests abnormal stress responses that are similar to other addictions (Geisel, Panneck, Hellweg, Wiedemann, & Muller, 2015; Paris, Franco, Sodano, Frye, & Wulfert, 2010), suggesting that a pattern of blunted stress reactivity may play a role in the behavioral changes that accompany gambling disorder. This work has not yet examined the mechanisms that link abnormal stress physiology with the behavioral disturbances of gambling disorder: namely the pattern of risky decision making that facilitates the development and maintenance of the disorder (van Holst, van den Brink, Veltman, & Goudriaan, 2010).
Stress and Gambling Disorder. Exaggerated cardiovascular and neuroendocrine reactivity
has well-documented negative health consequences (Chida & Steptoe, 2009; Lovallo & Gerin, 2003). Recent work suggests that reduced reactivity to stress may also serve as a marker for negative health outcomes, including addiction (Carroll, Ginty, Whittaker, Lovallo, & de Rooij, 2017; Miller, Chen, & Zhou, 2007; Phillips, Ginty, & Hughes, 2013). A vigorous stress response may serve as a marker of a healthy, adaptive physiological system, which may serve as a
diurnal dynamics of the hypothalamic pituitary adrenocortical (HPA) axis (al'Absi, 2018; Carroll et al., 2017; Lovallo, 2006; Meyer et al., 2004; Sinha, 2011).
Gambling disorder is now classified in the DSM-5 as a ÔSubstance-Related and Addictive DisorderÕ (American Psychiatric Association, 2013). This reflects an evolving understanding of the disorder, but a great deal remains unknown about gambling disorder. During a gambling episode, healthy individuals typically show a physiological response characterized by increased activity of the sympathetic nervous system (SNS) and HPA axis (Krueger, Schedlowski, & Meyer, 2005; Meyer et al., 2000; Meyer et al., 2004). These findings suggest that SNS and HPA physiology during gambling episodes reflect the arousing, but also potentially stressful, aspects of gambling. Such physiological activity may provide a bodily signal of stress that normally triggers cognitive and emotional regulatory processes to stop gambling upon reaching a threshold of physiological response. This normal pattern may be disrupted in gambling disorder, such that these bodily signals are dampened or absent, resulting in a lack of recognition of the stress of the gambling situation and financial losses that inevitably occur. In support of this idea, individuals with gambling disorder showed a blunted cortisol response to watching gambling-related videos compared to recreational gamblers (Paris et al., 2010). Similarly, lower resting heart rate
(Schmidt, Mussel, & Hewig, 2013) and lower resting cortisol (van Honk, Schutter, Hermans, & Putman, 2003) in healthy populations are associated with more risky decision making,
suggesting a relationship between blunted stress physiology and increased risk taking that may extend to the general population.
minutes after awakening (Pruessner et al., 1997) and the diurnal fall, a reduction in cortisol secretion from the morning to the evening (Evans et al., 2011). Alterations in these dynamics are associated with higher self-reported stress, as well as poorer mental, physical, and cognitive outcomes (Adam et al., 2017; Chida & Steptoe, 2009; Evans et al., 2011). In particular, a flatter slope of the cortisol fall is associated with negative health outcomes, including cancer,
cardiovascular disease, depression, and obesity, among other disorders (Adam et al., 2017). Research has also shown associations between diurnal cortisol patterns and addictive behavior. Adolescents who started drinking at an earlier age showed lower levels of cortisol before and
during a laboratory stressor (Evans, Greaves-Lord, Euser, Franken, & Huizink, 2012). Since
adolescents who start drinking earlier are more likely to develop alcohol use disorder, these findings, among others ( Moss, Vanyukov, Yao, & Kirillova, 1999; Moss, Vanyukov, & Martin, 1995; Sorocco, Lovallo, Vincent, & Collins, 2006), suggest that lower diurnal cortisol levels may serve as a physiological marker for the development of addiction.
Decision Making in Gambling Disorder. The types of decisions involved in gambling can
be characterized as occurring under risk or under uncertainty (see De Groot & Thurik, 2018 for review). An example of decision making under risk is betting on a coin flip, in which the
outcome is unknown (heads or tails), but the probability of winning is known (50%). By contrast, decision making under uncertainty (sometimes termed ÔambiguityÕ) is more akin to betting on a slot machine in which neither the outcome nor the probability of winning is known. People with gambling disorder perform worse on decision making under both risk and uncertainty compared to a control group (Brevers et al., 2012).
In both types of decision making, the term Ôrisk perceptionÕ is used to characterize the evaluation of risk parameters of the decision, including possible rewards, possible losses, and their respective probabilities. People with gambling disorder show poor risk perception, expecting better outcomes associated with gambling compared to non-gamblers (Spurrier & Blaszczynski, 2014). For example, individuals with gambling disorder are less attentive to monetary losses, but are highly responsive to greater monetary gains, suggesting heightened reward sensitivity (Boog, Hšppener, Goudriaan, Boog, & Franken, 2014; Brevers, Koritzky, Bechara, & No‘l, 2014; Marmurek, Switzer, & D'alvise, 2014). Poor risk perception among those with gambling disorder may lead to higher levels of risk taking behavior (e.g., greater impulsivity, steeper discounting rates, poor time perception, chasing propensity), all of which are associated with gambling severity (Ciccarelli, Griffiths, Nigro, & Cosenza, 2017; Madden, Petry, & Johnson, 2009; Nigro, Ciccarelli, & Cosenza, 2018).
out to examine these parameters in individuals with problem gambling1 compared to a healthy control group, and to assess how stress physiology may impact decision making under risk. We hypothesized that those with problem gambling would show a blunted cortisol awakening response and a flatter diurnal cortisol slope than the control group. In a set of exploratory analyses, we examined the relation between diurnal cortisol parameters and decision making under risk in a battery of gambling-type tasks that assess sensitivity to gains, losses, and expected value, both in terms of gambling behavior and self-reports of risk perception behind such behavior.
Methods
Participants
Community dwelling participants were recruited via advertisements in local newspapers, via Facebook, and flyers placed in community centers and retail establishments. Two groups of participants were recruited: (a) non-treatment seeking individuals with problem gambling (n = 30) and (b) healthy control participants (n = 29). Additional demographic data are provided in Table 1.
Inclusion criteria for all participants were being 21 years of age or older and English speaking. Exclusion criteria for all participants included uncontrolled serious psychiatric disorders (e.g., schizophrenia, manic episodes), current use of corticosteroid medications,
1Note that we use the term Ôproblem gamblingÕ to refer to our sample, which is
hormonal contraceptive use or current breastfeeding in women, positive breathalyzer test for alcohol intoxication, and positive urine toxicology tests for recent cocaine, opiates, or
amphetamines use. Inclusion criteria for the problem gambling group included report of at least four gambling episodes in the past 60 days, wagering at least $100 total, and reporting no recent treatment for gambling disorder. Inclusion criteria for healthy controls included no lifetime history of gambling disorder or substance dependence and no past-year history of mood or anxiety disorder.
An a priori power analysis was conducted based on previous data collected by our group. This analysis showed a difference in decision making on a probability discounting task between a group with gambling disorder and a control group of d = .75. A sample size of at least n = 23 per group is required to detect a statistically reliable group difference at 80% power. Our sample size of n = 30 was chosen to ensure adequate statistical power in detecting group differences in decision making.
Procedure
The study was approved by Saint Louis UniversityÕs Institutional Review Board. The study consisted of three phases: (a) Day 1: initial drug and alcohol screening, (b) Days 2 & 3: diurnal salivary cortisol collection over two consecutive days, and (c) Day 4: behavioral assessment in the laboratory. Participants were given a $100 ClinCard (Greenphire, King of Prussia, PA) for completing participation.
Drug and Alcohol Screening
Drug Test; Instant Technologies, Norfolk, VA). The drug and alcohol screenings were repeated before the behavioral assessments, two days later (see below), to confirm compliance with inclusion criteria. Participants who tested positive for alcohol intoxication, cocaine, opiates, or amphetamines on either day were excluded from the study; two participants were excluded and replaced for testing positive for drugs.
Gambling Severity
The following measures were used to determine whether a person met the clinical threshold for problem gambling.
The National Opinion Research Center DSM Screen for Gambling Problems (NODS)
(Gerstein et al., 1999). The NODS is a structured interview used to determine past-year and lifetime problem gambling. This measure consists of 17 questions reflecting the DSM-IV criteria and was devised by the National Opinion Research Center for the 1999 National Survey of Gambling Behavior; scoring for this measure was changed to reflect the new DSM-5 criteria. This measure has good internal consistency (α = .88; (Wickwire, Burke, Brown, Parker, & May, 2008). The problem gambling group in this study included individuals who scored a three or higher on the NODS; the healthy controls all scored zero on the screen. Although the threshold for problem gambling is lower than previously proposed for gambling disorder (i.e., four criteria), recent research suggests significant psychosocial impairment among subclinical gamblers who endorse fewer gambling disorder criteria (Weinstock, April, & Kallmi, 2017).
Gambling Timeline Followback (GTLFB). The GTLFB assesses the dollar amount wagered
gambling to spend at least $100 on 4 occasions over the previous 60 days. As reported by Weinstock and colleagues (2004), the GTLFB had good test-retest reliability among frequent (r = .75 - .96) and disordered gamblers (r = .73 - .93) as well as validity due to strong correlations between the measure and daily self-monitoring of gambling behavior.
Salivary Cortisol Collection
Participants collected saliva samples using Salivette collection devices (Sarstedt,
NŸmbrecht, Germany) at 10 time points over two consecutive weekdays outside the laboratory to reliably assess the cortisol awakening response and the post-awakening diurnal fall of cortisol (Hellhammer et al., 2007; Smyth, Thorn, Hucklebridge, Evans, & Clow, 2015). Sampling times included (a) awakening; (b) thirty minutes after awakening; (c) forty-five minutes after
awakening; (d) three hours after awakening; and (e) bedtime. For each participant, samples at each time point from day 1 were averaged with samples from the same time point on day 2. Cortisol awakening response was measured by calculating the area under the curve with respect to increase (AUCi; Pruessner, Kirschbaum, Meinlschmid, & Hellhammer, 2003) for the first three samples of the day. Diurnal cortisol slope was assessed by calculating the difference
between cortisol at bedtime and the morning peak (i.e., 45 minutes after waking) and dividing by the time awake (i.e., bedtime Ð peak / hours awake; Adam et al., 2017).
Recent work has shown that non-compliance with sampling instructions reduces reliability of diurnal cortisol dynamics measures; objective documentation of wake time using actigraphy and exact sampling time using objective monitoring are required for reliable assessment of cortisol dynamics (Smyth et al., 2015; Stalder et al., 2016). To document
when caps were removed. The cotton pledgets, normally stored in the Salivette device, were stored in the MEMS device to provide an objective time stamp of each sample. Also, FitBit activity monitors (FitBit Inc., San Francisco, California) were used to document wake times. The FitBit device shows high reliability and consistency with more traditional research-grade activity monitoring devices (Gusmer, Bosch, Watkins, Ostrem, & Dengel, 2014).
Participants were shown exactly how to complete saliva collection, and a research assistant was available via SMS messaging if participants had questions during the two days of sampling. Participants also received written instructions detailing the process, including: (a) instructions to wear the FitBit to bed the night after the first lab visit to objectively document sleep and wake time; (b) instructions on the exact times to collect saliva samples on both subsequent days; (c) instructions to collect saliva by opening the MEMS cap to remove one Salivette pledget at the time of each saliva sample; and (d) instructions to place the pledget in the Salivette device and stored it in participantsÕ home freezers prior to returning them to the
laboratory.
After the sampling days, participants returned the Salivette tubes, MEMS cap, and FitBit during the second laboratory visit. Saliva samples were stored at −20◦C until assayed.
Measurement of salivary cortisol was conducted via chemiluminescence assay (CLIA; IBL Hamburg, Germany) in the laboratory of Dr. Clemens Kirschbaum, Technical University of Dresden, Germany. Intraassay coefficients of variation were less than 10%. The lower sensitivity for cortisol is 0.5 nmol/l.
Behavioral Assessment
intoxication or recent drug use were excluded from further participation. Monetary incentives for all decision making tasks were hypothetical.
The Cups Task (Levin & Hart, 2003; Weller, Levin, Shiv, & Bechara, 2007). In the Cups
Task, participants made 54 choices between a certain and uncertain option, comprised of gain trials (in which money can be won) and loss trials (in which money can be lost). For gain trials, choosing a cup (displayed on a computer screen) from the Òcertain sideÓ always results in the gain of a dollar. The alternative is the Òrisky sideÓ in which participants are presented with the possibility to win more money Ð or win nothing. For each trial, they were informed whether two, three, or five dollars may be gained by choosing the ÒwinningÓ cup from a risky side array of two (50% chance to win), three (33%) or five (20%) cups. The three outcome magnitudes (i.e., number of dollars at stake) and three probabilities levels (i.e., number of cups) yield nine combinations, each iteration repeated three times in a random order, for a total of 27 reward trials.
decision-making (i.e., the tendency to make choices that have a more favorable long-term expected payout).
The Columbia Card Task (Figner, Mackinlay, Wilkening, & Weber, 2009) is a dynamic
computerized card game that assesses risk taking levels and information use strategies.
At the end of each version of the CCT, participants were asked to report what decision strategy they used to make their choices, with their responses to the following items: (a) I mainly focused on the number of loss cards; (b) I mainly focused on the gain amount; and (c) I mainly focused on the loss amount. Participants rated each of these options using a visual analogue scale (scored 0-100). These ratings were used to assess risk perception and whether those with
problem gambling used different information in making their choices than did the control group.
Statistical Analyses
Demographic characteristics (age, education, income, etc.) were analyzed between groups using t-tests for continuous data and chi-square for categorical data (see Table 1). Descriptions of analyses of cortisol and behavioral data are described in detail in the respective results sections.
Data Availability
Our data, as well as our code for statistical analyses, are publicly accessible at
https://osf.io/73h8q/. The stimulus materials may be obtained from the authors of the original tasks (the Cups Task: Levin & Hart, 2003 and the Columbia Card Task: Figner et al., 2009)).
Results
Diurnal Cortisol
Saliva Sampling Compliance. Missing samples were replaced with the value of the
followed the guidelines from Smyth and colleagues (2015), which indicate that cortisol data collected greater than 5 minutes outside of the expected sample times may result in inaccurate assessment of the cortisol awakening response. Compliance for these samples were measured by comparing the time from awakening (assessed via actigraphy) to the first opening of the MEMS cap. Based on this five-minute cut-off, 75% of gamblers were compliant and 77% of controls were compliant in their first saliva samples on each day; an independent t-test indicated no significant difference in compliance for the first sample between problem gambling (M[SD] = 6.41 [8.13] minutes from wake time) and control groups (M[SD] = 3.89 [4.74] minutes from wake time; t = 1.203, p = .237, d = 0.38). However, there were significant differences between groups for compliance on the 30-minute sample with higher non-compliance in the problem gambling (M[SD] = 16.09 [18.53] minutes from 30 min sample) compared to the control group (M[SD] = 4.50 [4.28] minutes from 30 min sample; t = 2.846, p = .026, d = 0.86). Despite this difference, the magnitude of the cortisol awakening response was not related to compliance of the wake (r = -.088, p > .05) nor the 30-minute sample (r = -.197, p > .05) among the problem gambling group, nor in the control group (wake: r = -.401, p > .05; 30-minute sample: r = -.279,
p > .05).
Cortisol Levels between Groups. A repeated measures ANOVA examining the time
course of the diurnal cycle between groups indicated a significant main effect of Group (F(1,57) = 5.392, p = .024, η2 = .086), and a Group ! Time interaction (F(2.86, 162.77) = 5.81; p = .001;
(F(2.86, 162.77) = 120.39; p < .001; η2 = .68; see Figure 1). MauchlyÕs test of sphericity was significant and therefore a Greenhouse-Geisser correction was used.
Next, we assessed individual components of the diurnal cycle between groups, the cortisol awakening response (CAR) and the diurnal slope. An independent samples t-test indicated no significant differences in the CAR between the problem gambling (M[SD] = 3.39 [6.61]) and control group (M[SD] = 6.62 [7.13]; t[57] = 1.81; p = .076; d = .47; 95% CI: 6.81 .35). However, there was a significant group difference in diurnal cortisol slope (t[57] = 3.12; p = .003; d = .86; 95% CI: .008 - .035) such that those with problem gambling had a flatter slope (M[SD] = -.036 [.028]) than the control group (M[SD] = -.058 [.023]). LeveneÕs tests indicated no issues of heterogeneity of variance for CAR or diurnal cortisol slope.
Correlation analyses were conducted to examine the relation between problem gambling severity and diurnal cortisol patterns, including only participants in the problem gambling group. Lower CAR was associated with higher NODS scores, r = -.396, p = .015; CAR was not
associated with any GTLFB variables. We found no significant relations between cortisol slope and gambling severity indices (i.e., NODS, GTLFB). Altogether, these data demonstrate blunted cortisol levels throughout the morning, which translates to a flatter diurnal cortisol slope in those with problem gambling. Further, within the problem gambling group, lower CAR was associated with greater gambling severity, as assessed with the NODS, suggesting that these alterations are related to problem gambling pathology.
Behavioral Data
Columbia Card Task (CCT). For the CCT analyses, we used a linear mixed-effects model
intercept, a fixed effect for the ÒhotÓ versus ÒcoldÓ CCT factor (all factors were coded using sum-to-zero contrasts), fixed effects for the group factor, the three card game factors probability, gain amount, and loss amount, as well as a standardized (i.e., centered and scaled) continuous predictor for Block (indicating the three blocks of eight trials each). Additionally, we included fixed effects for the two-way interaction terms involving condition by task factors, as well as those involving group by task factors. We also included the three-way interactions among condition, group, and each of the card game factors.
We followed Barr, Levy, Scheepers, and Tily's (2013) recommendation to use a maximal random-effects structure: The repeated-measures nature of the data was accordingly modeled by including a per-participant random adjustment to the fixed intercept (Òrandom interceptÓ), as well as per-participant random adjustments to all within-subject predictors (i.e., condition, probability, gain, loss, block, and the two-way interactions between condition and each of the card game factors) (Barr, Levy, Scheepers, & Tily, 2013). The random covariance terms were modeled with an autoregressive structure among the random effects. Significance was
determined by calculating 95% confidence intervals for the parameter estimates using bootstrapping with 2000 resamples.
Table 2 displays the parameter estimates for the CCT mixed-effects models analysis. At the task level, we found a main effect for task version; thus, participants were more likely to turn over more cards on the CCT-Cold version than for the CCT-Hot. Additionally, we found main effects for the probability, gain magnitude, and loss magnitude CCT primitive factors.
average. As they learn to reduce this number, they typically will earn more points. Central to our hypotheses, we found that the problem gambling group turned over more cards on the CCT than did the control group, regardless of condition.
These effects were conditional on several interaction effects. We found significant two-way interactions at the task level. Specifically, CCT Condition interacted with (a) loss magnitude and (b) gain magnitude, with each effect suggesting greater information use (i.e., sensitivity to changes in magnitude) in the Cold version. Central to our study, we also observed significant two-way interactions between group and the task characteristics, holding other variables
constant. Though the problem gambling group took more risks overall, they were especially apt to do so in the Cold condition (see Figure 2). Additionally, they were less sensitive to changes in number of loss cards from 1 to 3 (i.e., probability level). Further, the gambling group was less likely to adjust to a more risk-averse strategy like the control group did, evidenced by a significant Group by Block interaction. The tested three-way interactions were not significant.
CCT Strategy Use. The problem gambling group indicated paying more attention to the
gain points than did the healthy controls on both the CCT-Hot (t[57] = 2.55, p = .017, 95% CI: 4.20 Ð 35.21, d = .68) and CCT-Cold (t[57] = 2.38, p = .021, 95% CI: 2.89 Ð 33.65, d = .63; see Table 3). There were no significant differences between groups on self-reported attention paid to number of loss cards or loss amount. Combined, these results from behavior and self-reported strategy use on the CCT, indicate that those with problem gambling made more risky choices, did not alter their behavior over the course of the task, and reported focusing more on gains than did the control group.
Cups Task. Analyses for the Cups Task were similar to those for the CCT, with the
used a logit-link function for a generalized linear mixed model analysis framework. The main omnibus model included a fixed intercept, a fixed effect for domain (gain/loss; all factors were coded using sum-to-zero contrasts), fixed effects for the problem gambling group factor, and the two task factors: probability and outcome magnitude. Additionally, we included fixed effects for the two-way interactions between group and each of the task factors, as well as domain and the task factors. Finally, we included, the three-way interactions among condition, group, and each of the card game factors. The repeated-measures nature of the data was accordingly modeled by including a random intercept, as well as per-participant random adjustments to all within-subject predictors (i.e., domain, probability, outcome, and the two-way interactions between domain and both the task factors). The random covariance terms were modeled with an autoregressive structure among the random effects.
Consistent with expectations due to the Cups Task payout structure, we found significant domain by probability and domain by outcome interactions (see Table 4). Specifically,
individuals took fewer risks as the number of Cups changed from 2 (50% probability) to 5 (20%) cups for the gain domain, and took fewer risks as the number of cups decreased in the loss domain. Similarly, this pattern was apparent for outcome magnitude; increases in the potential risky gain related to greater risk-taking, whereas increases in the amount that one could lose related to less risk-taking. Holding these effects constant, though, we also found an effect for group, in which the problem gambling group made more risky choices than did the control group, regardless of whether or not the decision was presented as a potential gain or potential loss.
Given the robust findings suggesting differences between the problem gambling and control groups, we tested the degree to which diurnal cortisol dynamics were associated with performance on the CCT and Cups task in the problem gambling group. To do this, we modified the mixed models described above to include gamblersÕ CAR and diurnal slope as fixed effects, also allowing these variables to interact with task factors.
CCT. As shown in Table 5, we found that a more blunted CAR was associated with
increased risk-taking on the CCT. This effect, however, is conditional on a CAR by Condition interaction. As shown in Figure 3, although lower CAR was associated with increased overall risk taking, this effect was most pronounced on the Hot version of the CCT. We did not find any significant effects with respect to the diurnal fall indicator.
Cups Task. Table 6 reports the parameter estimates for the generalized linear mixed
model analysis for the Cups task. In contrast to the CCT, we did not find any significant effects for either CAR or diurnal fall, either in terms of main effects or their interactions with the task factors.
Findings from these analyses demonstrate a potential influence of blunted morning cortisol on specific aspects of decision making among individuals with problem gambling.
Discussion
Second, compared to control individuals, participants with problem gambling showed elevated risk-taking preferences on two often-used risky decision-making tasks. We found evidence that this elevated pattern was due to the problem gambling group showing lower sensitivity to changes in probability across trials, showing more optimism than controls when there was a low probability for a loss, and paying more attention to gains than either losses or probabilities. Third, we observed that for the problem gambling group, lower cortisol was associated with more risk taking, especially when the decision context was more affective in nature (in the Hot versus the Cold CCT).
Compared to an age- and sex-matched control group, individuals with problem gambling showed lower cortisol levels at both 30 and 45 minutes after awakening, resulting in a flatter cortisol slope. The groups did not differ in cortisol levels at wake time, in the afternoon, nor at bedtime. These findings are consistent with studies demonstrating a blunted stress response in addiction populations (Lovallo, 2006). Such blunted responsivity may reflect a physiological phenotype of addiction, such that individuals with lowered physiological reactivity to stress, and in this case, to awakening, may be more likely to engage in addictive behaviors or show blunted reactivity due to the addictive behavior. No matter the pathway, these reduced bodily signals may lead to greater sensation seeking among at-risk populations that ultimately play a role in
between the current study and that of Wohl and colleagues (2008). First, our problem gambling population was considerably older and reported more severe problem gambling than the first-year college student population assessed in the previous study. Second, in the previous study, saliva samples were collected on only one day and did not include objective measures to determine the timing of awakening or saliva sampling, which may have resulted in imprecise measures of the cortisol awakening response. These relations between diurnal cortisol and gambling behavior should be examined in more problem gambling populations to more fully examine the relationship between these variables.
People in the problem gambling group made more risky choices in both the hot and the cold versions of the CCT. These results are in line with our hypothesis and previous work documenting increased risk preferences among those with problem gambling (Giorgetta et al., 2014; Ring et al., 2018; Takeuchi et al., 2016). Previous research showed that adolescents performing better on a measure of inhibitory control (Go/No-Go task) took fewer risks on the CCT-hot (Figner & Weber, 2011). We did not assess inhibitory control in our participants, but previous work with gambling disorder populations shows impaired inhibition performance (Grant & Chamberlain, 2014). These findings, coupled with the relationship between adolescent inhibitory control and CCT-hot performance, suggest that inhibitory control may play a role in the increased risk taking behavior observed in our sample.
and/or responses to stress pre-date the elevated risk taking in this population. Among those with problem gambling, lower cortisol was also associated with more optimistic choices in decision making under risk, both on the hot and cold versions of the CCT, regardless of whether the risk was presented as a potential gain or loss. When presented with trials with a high loss amount (750 points), most participants turn over fewer cards, compared to trials with a low loss amount (250 points). Those problem gambling participants with lower cortisol were less likely to change their behavior based on potential loss amount. This finding is consistent with recent work
showing that problem gamblers (who fulfilled at least three of the DSM-IV criteria for
pathological gambling) were not sensitive to changes in the loss domain, but focused on changes in the gain domain only (Ring et al., 2018). These findings are consistent with the probability distortion hypothesis, the idea that problem gambling is related to an overweighting of small probabilities of winning, as well as the probability elevation hypothesis, or a general
overweighting of gain probabilities (Ligneul, Sescousse, Barbalat, Domenech, & Dreher, 2013). It also ties into maladaptive cognitive distortions endorsed by individuals with problem
gambling. One cognitive distortion frequently endorsed by gamblers is the Òavailability
illusionÓ, in which an individual more easily recalls times when they won versus the much more common experience of losing(Ciccarelli et al., 2017). Severity of gambling disorder, assessed with the South Oaks Gambling Screen is positively correlated with a measure of Ôrisk
attractivenessÕ in a gambling disorder population (Ligneul et al., 2013). Our results
demonstrating a relation between blunted CAR, higher problem gambling severity assessed with the NODS, and more optimistic choices in the CCT leads to the speculation that lower cortisol levels may serve as a marker for gambling severity, which may be addressed in future
for substance use disorders, such as individuals with a family history of the disorder (see Lovallo, 2006 for review). The similarities between gambling disorder and other addictions further support speculation about the relationship between blunted cortisol and gambling disorder.
Analysis of self-reported decision strategy about what information participants attended to in the CCT showed that individuals with problem gambling reported paying more attention to gain information (i.e., number of points gained per gain card) compared to the control group. These findings extend previous work on behavioral choice to subjective report and demonstrate that those with problem gambling have some insight into the motivation behind their behavior. The groups did not differ on attention paid to loss information (i.e., number of points lost when a loss card was chosen) or the probability of picking a loss card. This finding is consistent with our behavioral results and previous work demonstrating a biased focus on gains in problem gambling (Ring et al., 2018).
shown that the Cups Task is sensitive to group-level differences between vulnerable groups that have been associated with greater levels of psychological dysregulation (Weller et al., 2007; Weller, Levin, & Bechara, 2010; Weller, Kim, Leve,Bhimji, & Fisher, 2015). Thus, future research with a larger sample is warranted.
A few limitations of our study should be noted. We classified our participants in a binary, rather than a dimensional approach that takes problem gambling severity into account (e.g., Canadian Problem Gambling Severity Index classifications; (Holtgraves, 2009)). This
classification scheme does not allow for detailed problem gambling severity to be accounted for in our statistical models. Despite this, severity of problem gambling in our sample, assessed with the NODS, was associated with lower cortisol levels. Our relatively low sample size (another limitation of the study) precluded further subdividing participants along a severity scale for the modeling analyses. Another limitation of the study is the use of laboratory tasks without real financial risks or real financial gains. Although performance on these tasks differentiated the problem gambling group from the control group, these tasks may not reflect real-world decision making.
References
Adam, E. K., Quinn, M. E., Tavernier, R., McQuillan, M. T., Dahlke, K. A., & Gilbert, K. E. (2017). Diurnal cortisol slopes and mental and physical health outcomes: A systematic review and meta-analysis. Psychoneuroendocrinology, 83, 25-41.
al'Absi, M. (2018). Stress and addiction: When a robust stress response indicates resiliency.
Psychosomatic Medicine, 80(1), 2-16.
American Psychiatric Association, A. P. (2013). Diagnostic and Statistical Manual of Mental
Disorders (5th Edition ed.). Washington, D.C.: American Psychiatric Association.
Barr, D. J., Levy, R., Scheepers, C., & Tily, H. J. (2013). Random effects structure for
confirmatory hypothesis testing: Keep it maximal. Journal of Memory and Language,
68(3), 255-278.
Biback, C., & Zack, M. (2015). The Relationship Between Stress and Motivation in Pathological Gambling: a Focused Review and Analysis. Current Addiction Reports, 2(3), 230-239. Boog, M., Hšppener, P., Goudriaan, A. E., Boog, M. C., & Franken, I. H. (2014). Cognitive
inflexibility in gamblers is primarily present in reward-related decision making. Frontiers
in Human Neuroscience, 8, 569.
Brevers, D., Cleeremans, A., Goudriaan, A. E., Bechara, A., Kornreich, C., Verbanck, P., & No‘l, X. (2012). Decision making under ambiguity but not under risk is related to problem gambling severity. Psychiatry Research, 200, 568-574.
Brevers, D., Koritzky, G., Bechara, A., & No‘l, X. (2014). Cognitive processes underlying impaired decision-making under uncertainty in gambling disorder. Addictive Behaviors,
Buchanan, T. W., & Lovallo, W. R. (2018). The role of genetics in stress effects on health and addiction. Current Opinion in Psychology, 27, 72-76.
Carroll, D., Ginty, A. T., Whittaker, A. C., Lovallo, W. R., & de Rooij, S. R. (2017). The behavioural, cognitive, and neural corollaries of blunted cardiovascular and cortisol reactions to acute psychological stress. Neuroscience & Biobehavioral Review, 77, 74-86.
Chida, Y., & Steptoe, A. (2009). Cortisol awakening response and psychosocial factors: a systematic review and meta-analysis. Biological Psychology, 80, 265-278.
Ciccarelli, M., Griffiths, M. D., Nigro, G., & Cosenza, M. (2017). Decision making, cognitive distortions and emotional distress: a comparison between pathological gamblers and healthy controls. Journal of Behavior Therapy and Experimental Psychiatry, 54, 204-210.
De Groot, K., & Thurik, R. (2018). Disentangling risk and uncertainty: When risk-taking measures are not about risk. Frontiers in Psychology, 9, 2194.
Evans, B. E., Greaves Lord, K., Euser, A. S., Franken, I. H., & Huizink, A. C. (2012). The relation between hypothalamic pituitary adrenal (HPA) axis activity and age of onset of alcohol use. Addiction, 107, 312-322.
Evans, P. D., Fredhoi, C., Loveday, C., Hucklebridge, F., Aitchison, E., Forte, D., & Clow, A. (2011). The diurnal cortisol cycle and cognitive performance in the healthy old.
Figner, B., Mackinlay, R. J., Wilkening, F., & Weber, E. U. (2009). Affective and deliberative processes in risky choice: age differences in risk taking in the Columbia Card Task.
Journal of Experimental Psychology: Learning, Memory, and Cognition, 35, 709.
Figner, B., & Weber, E. U. (2011). Who takes risks when and why? Determinants of risk taking.
Current Directions in Psychological Science, 20, 211-216.
Geisel, O., Panneck, P., Hellweg, R., Wiedemann, K., & Muller, C. A. (2015). Hypothalamic-pituitary-adrenal axis activity in patients with pathological gambling and internet use disorder. Psychiatry Res, 226, 97-102.
Gerstein, D., Volberg, R. A., Toce, M., Harwood, H., Johnson, R., Buie, T., . . . Engelman, L. (1999). Gambling impact and behavior study: Report to the national gambling impact study commission. Chicago: National Opinion Research Center.
Giorgetta, C., Grecucci, A., Rattin, A., Guerreschi, C., Sanfey, A. G., & Bonini, N. (2014). To play or not to play: a personal dilemma in pathological gambling. Psychiatry Research,
219, 562-569.
Grant, J. E., & Chamberlain, S. R. (2014). Impulsive action and impulsive choice across substance and behavioral addictions: cause or consequence? Addictive Behaviors, 39, 1632-1639.
Gusmer, R., Bosch, T., Watkins, A., Ostrem, J., & Dengel, D. (2014). Comparison of FitBit¨ Ultra to ActiGraphª GT1M for Assessment of Physical Activity in Young Adults During Treadmill Walking. The Open Sports Medicine Journal, 8(1).
Holtgraves, T. (2009). Evaluating the problem gambling severity index. Journal of Gambling
Studies, 25(1), 105-120.
Krueger, T. H., Schedlowski, M., & Meyer, G. (2005). Cortisol and heart rate measures during casino gambling in relation to impulsivity. Neuropsychobiology, 52, 206-211.
Levin, I. P., & Hart, S. S. (2003). Risk preferences in young children: Early evidence of individual differences in reaction to potential gains and losses. Journal of Behavioral
Decision Making, 16, 397-413.
Ligneul, R., Sescousse, G., Barbalat, G., Domenech, P., & Dreher, J.-C. (2013). Shifted risk preferences in pathological gambling. Psychological Medicine, 43, 1059-1068.
Lovallo, W. R. (2006). Cortisol secretion patterns in addiction and addiction risk. International
Journal of Psychophysiology, 59, 195-202.
Lovallo, W. R., & Gerin, W. (2003). Psychophysiological reactivity: mechanisms and pathways to cardiovascular disease. Psychosomatic Medicine, 65, 36-45.
Madden, G. J., Petry, N. M., & Johnson, P. S. (2009). Pathological gamblers discount probabilistic rewards less steeply than matched controls. Experimental and Clinical
Psychopharmacology, 17, 283.
Marmurek, H. H., Switzer, J., & D'alvise, J. (2014). A comparison of university student and community gamblers: Motivations, impulsivity, and gambling cognitions. Journal of
Behavioral Addictions, 3, 54-64.
Meyer, G., Hauffa, B. P., Schedlowski, M., Pawlak, C., Stadler, M. A., & Exton, M. S. (2000). Casino gambling increases heart rate and salivary cortisol in regular gamblers. Biological
Meyer, G., Schwertfeger, J., Exton, M. S., Janssen, O. E., Knapp, W., Stadler, M. A., . . . KrŸger, T. H. (2004). Neuroendocrine response to casino gambling in problem gamblers.
Psychoneuroendocrinology, 29, 1272-1280.
Miller, G. E., Chen, E., & Zhou, E. S. (2007). If it goes up, must it come down? Chronic stress and the hypothalamic-pituitary-adrenocortical axis in humans. Psychological Bulletin,
133(1), 25-45.
Moss, H. B., Vanyukov, M., Yao, J. K., & Kirillova, G. P. (1999). Salivary cortisol responses in prepubertal boys: the effects of parental substance abuse and association with drug use behavior during adolescence. Biological Psychiatry, 45, 1293-1299.
Moss, H. B., Vanyukov, M. M., & Martin, C. S. (1995). Salivary cortisol responses and the risk for substance abuse in prepubertal boys. Biological Psychiatry, 38, 547-555.
Nigro, G., Ciccarelli, M., & Cosenza, M. (2018). The illusion of handy wins: Problem gambling, chasing, and affective decision-making. Journal of Affective Disorders, 225, 256-259. Paris, J., Franco, C., Sodano, R., Frye, C., & Wulfert, E. (2010). Gambling pathology is
associated with dampened cortisol response among men and women. Physiology &
Behavior, 99, 230-233.
Petry, N. M., Stinson, F. S., & Grant, B. F. (2005). Comorbidity of DSM-IV pathological gambling and other psychiatric disorders: results from the National Epidemiologic Survey on Alcohol and Related Conditions. Journal of Clinical Psychiatry, 66, 564-574. Phillips, A. C., Ginty, A. T., & Hughes, B. M. (2013). The other side of the coin: blunted
cardiovascular and cortisol reactivity are associated with negative health outcomes.
International Journal of Psychophysiology, 90, 1-7.
Pruessner, J. C., Kirschbaum, C., Meinlschmid, G., & Hellhammer, D. H. (2003). Two formulas for computation of the area under the curve represent measures of total hormone
concentration versus time-dependent change. Psychoneuroendocrinology, 28, 916-931. Pruessner, J. C., Wolf, O. T., Hellhammer, D. H., Buske-Kirschbaum, A., von Auer, K., Jobst,
S., . . . Kirschbaum, C. (1997). Free cortisol levels after awakening: a reliable biological marker for the assessment of adrenocortical activity. Life Sciences, 61, 2539-2549. Ring, P., Probst, C. C., Neyse, L., Wolff, S., Kaernbach, C., van Eimeren, T., . . . Schmidt, U.
(2018). ItÕs all about gains: Risk preferences in problem gambling. Journal of
Experimental Psychology: General, 147, 1241-1255.
Schmidt, B., Mussel, P., & Hewig, J. (2013). I'm too calmÑLet's take a risk! On the impact of state and trait arousal on risk taking. Psychophysiology, 50, 498-503.
Sinha, R. (2011). New findings on biological factors predicting addiction relapse vulnerability.
Current Psychiatry Reports, 13, 398-405.
Smyth, N., Thorn, L., Hucklebridge, F., Evans, P., & Clow, A. (2015). Post awakening salivary cortisol secretion and trait well-being: The importance of sample timing accuracy.
Psychoneuroendocrinology, 58, 141-151.
Sorocco, K. H., Lovallo, W. R., Vincent, A. S., & Collins, F. L. (2006). Blunted hypothalamicÐ pituitaryÐadrenocortical axis responsivity to stress in persons with a family history of alcoholism. International Journal of Psychophysiology, 59, 210-217.
Spurrier, M., & Blaszczynski, A. (2014). Risk perception in gambling: A systematic review.
Stalder, T., Kirschbaum, C., Kudielka, B. M., Adam, E. K., Pruessner, J. C., Wust, S., . . . Clow, A. (2016). Assessment of the cortisol awakening response: Expert consensus guidelines.
Psychoneuroendocrinology, 63, 414-432.
Takeuchi, H., Kawada, R., Tsurumi, K., Yokoyama, N., Takemura, A., Murao, T., . . . Takahashi, H. (2016). Heterogeneity of loss aversion in pathological gambling. Journal of Gambling
Studies, 32, 1143-1154.
Tversky, A., & Kahneman, D. (1983). Extensional versus intuitive reasoning: The conjunction fallacy in probability judgment. Psychological Review, 90, 293.
van Holst, R. J., van den Brink, W., Veltman, D. J., & Goudriaan, A. E. (2010). Why gamblers fail to win: a review of cognitive and neuroimaging findings in pathological gambling.
Neuroscience & Biobehavioral Reviews, 34, 87-107.
van Honk, J., Schutter, D. J., Hermans, E. J., & Putman, P. (2003). Low cortisol levels and the balance between punishment sensitivity and reward dependency. Neuroreport, 14, 1993-1996.
Weinstock, J., April, L. M., & Kallmi, S. (2017). Is subclinical gambling really subclinical?
Addictive Behavior, 73, 185-191.
Weinstock, J., Whelan, J. P., & Meyers, A. W. (2004). Behavioral assessment of gambling: an application of the timeline followback method. Psychological Assessment, 16, 72-82. Weller, J. A., Buchanan, T. W., Shackleford, C., Morganstern, A., Hartman, J. J., Yuska, J., &
Weller, J. A., Kim, H., Leve, L.D., Bhimji, J., & Fisher, P.A. (2015). Plasticity of decision-making abilities for maltreated adolescents: evidence from a random clinical trial intervention. Development and Psychopathology, 27, 535-551.
Weller, J. A., Levin, I. P., & Bechara, A. (2010). Do individual differences in Iowa Gambling Task performance predict adaptive decision making for risky gains and losses? Journal of
Clinical and Experimental Neuropsychology, 32, 141Ð150.
Weller, J. A., Levin, I. P., Shiv, B. & Bechara, A. (2007). Neural correlates of adaptive decision making for risky gains and losses. Psychological Science, 18, 958Ð964.
Wickwire, E. M., Burke, R. S., Brown, S. A., Parker, J. D., & May, R. K. (2008). Psychometric evaluation of the national opinion research center DSM-IV screen for gambling problems (NODS). American Journal on Addictions, 17(5), 392-395.
Table 1. Demographic characteristics by group Variable Problem Gambling (n = 30) Healthy Control (n = 29) Statistic (df), p-value Gender Male Female 17 (56.7%) 13 (43.3%) 10 (33.3%) 20 (66.7%) χ2(1) = 3.30, p = .069 Marital Status Single Married Divorced/Separated Widower/Other 19 (63.3%) 7 (23.3%) 3 (10.0%) 1 (3.3%) 15 (51.7%) 8 (27.6%) 6 (20.6%) 0 (0.0%) χ2(4) = 2.81, p = .591 Race Caucasian African American Other 17 (56.7%) 12 (40.0%) 1 (3.3%) 23 (79.3%) 5 (17.2%) 1 (3.4%) χ2(3) = 5.77, p = .124 Ethnicity Hispanic Non-Hispanic 0 (0.0%) 30 (100%) 1 (3.4%) 28 (96.60%) χ2(1) = 1.05, p = .305 Employment Status Full Time Part Time Retired Disability Student Other 16 (53.3%) 3 (10.0%) 3 (10.0%) 3 (10.0%) 0 (0.0%) 5 (16.7%) 16 (55.2%) 2 (6.9%) 6 (20.7%) 0 (0.0%) 2 (6.9%) 3 (10.3%) χ2(5) = 6.69, p = .245 M (SD) M (SD) Age (years) 45.63 (12.75) 42.70 (13.81) t(52) = .809, p = .422 Education (years) 14.10 (2.04) 16.07 (4.23) t(57) = -2.29, p = .026
Table 2 Linear Mixed Models Analysis for CCT Performance in All Participants
Fixed Effects Parameter
estimate Std. Error 95% Confidence Interval Low High (intercept) 10.81 0.125 10.56 11.04 Group -1.61* 0.129 -1.85 -1.34 Condition (Hot/Cold) 1.84* 0.124 1.60 2.10 Block -0.30* 0.121 -0.54 -0.06 Probability -2.63* 0.122 -2.86 -2.40 Gain 0.25* 0.120 0.01 0.49 Loss -0.49* 0.120 -0.72 -0.26 Group X Condition -0.51* 0.126 -0.76 -0.27 Group X Block -0.30* 0.118 -0.54 -0.07 Group X Probability 0.34* 0.119 0.11 0.57 Group X Gain -0.11 0.120 -0.34 0.13 Group X Loss -0.14 0.122 -0.37 0.11 Condition X Block 0.03 0.121 -0.21 0.27 Condition X Probability 0.03 0.119 -0.20 0.27 Condition X Gain 0.28* 0.120 0.05 0.52 Condition X Loss -0.38* 0.120 -0.62 -0.15
Block X Condition X Group 0.08 0.119 -0.16 0.31
Probability X Condition X
Group -0.20 0.120 -0.43 0.04
Gain X Condition X Group -0.15 0.120 -0.38 0.09
Loss X Condition X Group 0.09 0.119 -0.13 0.33
Table 3. Ratings of Strategy Use During CCT Between Groups Means (SD) t df p Problem Gambling Control d 95% CI CCT-Hot Probability 57.07 (31.42) 48.46 (28.28) 1.14 54 .259 .305 -6.91 Ð 25.13 Gain 57.11 (28.96) 35.75 (28.23) 2.79 54 .007 .747 6.04 Ð 36.68 Loss 49.89 (32.44) 54.75 (32.37) .56 54 .577 .149 -22.22 Ð 12.51 CCT-Cold Probability 65.55 (27.87) 68.36 (31.79) -.35 55 .724 .094 -18.66 Ð 13.05 Gain 56.66 (32.01) 39.39 (26.55) 2.21 55 .031 .586 1.62 Ð 32.90 Loss 48.03 (33.37) 51.36 (32.97) -.38 55 .707 .100 -20.94 Ð 14.29
Table 4 Linear Mixed Models Analysis for Cups Task Performance in All Participants
Fixed Effects Parameter
estimate Std. Error 95% Confidence Interval Low High (intercept) .52 .166 .19 .84 Group -.34 .166 -.66 -.01 Domain (Gain/Loss) .04 .127 -.21 .29 Probability -.07 .063 -.20 .06 Outcome .17* .074 .02 .32 Group X Domain .09 .127 -.16 .34 Group X Probability .02 .063 -.11 .15 Group X Outcome -.11 .074 -.22 .08 Domain X Probability -.37* .071 -.52 -.23 Domain X Outcome -.45* .077 -.60 -.29 Group X Domain X Probability -.14 .071 -.28 .01
Group X Domain X Outcome -.04 .077 -.19 .12
Table 5 Linear Mixed Models Analysis: Cortisol Indicators Moderate CCT Task Performance
in Problem Gambling Group
95% CI Estimate Std. Error Lower Bound Upper Bound Intercept 12.27 0.184 11.8749 12.60 CAR -0.51* 0.223 -0.94 -0.07 Diurnal Fall 0.17 0.226 -0.27 0.63 Condition 2.36* 0.177 2.00 2.71 Block 0.01 0.175 -0.34 0.35 Probability -3.09* 0.169 -3.4 -2.76 Gain 0.35 0.178 -0.01 0.70 Loss -0.51* 0.171 -0.86 -0.18 CAR X Condition 0.53* 0.220 0.10 0.98 CAR X Block 0.26 0.210 -0.16 0.70 CAR X Probability 0.06 0.205 -0.33 0.47 CAR X Gain -0.21 0.207 -0.63 0.21 CAR X Loss -0.32 0.208 -0.74 0.09
Diurnal Fall X Condition 0.21 0.224 -0.21 0.66
Diurnal Fall X Block 0.10 0.210 -0.31 0.50
Diurnal Fall X Probability 0.38 0.208 -0.02 0.79
Diurnal Fall X Gain -0.06 0.213 -0.49 0.38
Diurnal Fall X Loss 0.30 0.205 -0.10 0.71
Condition X Block -0.07 0.176 -0.41 0.27
Condition X Probability 0.15 0.168 -0.16 0.49
Condition X Gain 0.47* 0.171 0.13 0.82
Condition X Loss -0.61* 0.175 -0.96 -0.26
CAR X Condition X Block -0.26 0.218 -0.70 0.18
CAR X Condition X Probability -0.35 0.210 -0.76 0.06
CAR X Condition X Gain -0.25 0.214 -0.67 0.17
CAR X Condition X Loss -0.18 0.212 -0.59 0.24
Diurnal Fall X Condition X Block -0.06 0.210 -0.47 0.33
Diurnal Fall X Condition X Probability 0.04 0.210 -0.38 0.43
Diurnal Fall X Condition X Gain -0.25 0.204 -0.66 0.16
Diurnal Fall X Condition X Loss 0.31 0.209 -0.09 0.72
Table 6: Linear Mixed Models Analysis: Cortisol Indicators and Cups Task Performance in
Problem Gambling Group
95% CI Estimate Std. Error Lower Bound Upper Bound Intercept 0.84 0.142 0.56 1.13 CAR 0.06 0.147 -0.24 0.35 Diurnal Fall 0.02 0.140 -0.26 0.30 Domain 0.04 0.142 -0.24 0.32 Probability -0.15 0.150 -0.44 0.15 Outcome 0.24 0.151 -0.06 0.54 CAR X Domain 0.00 0.147 -0.29 0.30 CAR X Probability -0.09 0.155 -0.40 0.22 CAR X Outcome -0.04 0.156 -0.35 0.27
Diurnal Fall X Domain -0.18 0.140 -0.46 0.10
Diurnal Fall X Probability 0.09 0.149 -0.21 0.38
Diurnal Fall X Outcome -0.08 0.149 -0.38 0.22
Domain X Probability -0.20 0.150 -0.50 0.10
Domain X Outcome -0.43* 0.151 -0.73 -0.13
CAR X Domain X Probability -0.02 0.155 -0.29 0.33
CAR X Domain X Outcome 0.02 0.156 -0.29 0.33
Diurnal Fall X Domain X Probability -0.06 0.149 -0.36 0.23
Diurnal Fall X Domain X Outcome 0.10 0.149 -0.19 0.40
Figure Legends
Figure 1: Mean cortisol levels (+ SEM) after awakening across groups. The problem gambling group showed blunted cortisol levels at 30 and 45 min after awakening and a flatter diurnal slope; W = wake.
Figure 2. Mean (+ SEM) number of cards turned in the Columbia Card Task across groups. The problem gambling group showed greater risk taking across both the hot and the cold versions of the CCT.
